Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Exclusion restriction with mprobit

From   saqlain raza <>
To   "" <>
Subject   Re: st: Exclusion restriction with mprobit
Date   Thu, 3 Nov 2011 14:20:58 -0700 (PDT)

 I am using two binary dependent variables and a set of independent covariates. For my case, the errors are correlated with each other. What do you suggest?

Saqlain RAZA
From: Maarten Buis <>
Sent: Wednesday, November 2, 2011 9:25 PM
Subject: Re: st: Exclusion restriction with mprobit

On Wed, Nov 2, 2011 at 9:06 PM, saqlain raza wrote:
> I am trying to use "mprobit" to estimate a multinomial probit model with four alternatives.
> As far as I know, what we need for identifying MNP model is some set of proper "exclusion
> restrictions," (as illustrated by Keane (1992)), not the suggested format of the data itself.

That is only necesary when your multinomial probit model includes
correlations between the different error terms. The -mprobit- command
in Stata sets these correlations at 0, imposing the equivalent of the
IAA assumption to a multinomial probit model. It is this assumption
that identifies the model rather than an exclusion restriction.

Hope this helps,

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index